Week 5: Land Cover Change Modeling

Yi Qiang
Feb. 10, 2017

Space and Time



Give me space and motion and I will give you a world.


R Decartes

Land (Use) Cover Change

  • Biophysical materials and human-made features are dynamic, changing rapidly.
  • An important instrument to study interactions between natural and human systems on the earth surface.
  • An important indicator and factor to climate change.
  • Significant effort has done in the development of change detection methods using remotely sensed data.

General Steps

1. Data processing

  • Making images comparable (Rectification, classification, interpolation, resampling)

2. Change Detection (Exploratory analysis)

  • How much and where were the changes

3. Change Modeling (Empirical modeling)

  • How it changed and what caused the change

4. Change Simulation (Predictive modeling)

  • How it will change in future or in other scenarios?

1. Data Processing

  • Successful remote sensing change detection requires careful attention to:
    • remote sensor system considerations, and
    • environmental characteristics.
  • Ideally, the remotely sensed data used to perform change detection is acquired by a remote sensor system that holds the following resolutions constant: temporal, spatial (and look angle), spectral, and radiometric.
  • Many processed land cover databases are available, e.g. (NLCD, NOAA C-CAP)

1. Data Processing

Spatial Resolution

  • Land cover change detection should be conducted among images with same spatial resolution and grid

  • If not, images need to be resampled or interplated into the same resolution and grid - but this process will introduce error or information loss.

1. Data Processing

Temporal Resolution

  • Be careful to identify the optimal change detection time period(s).

    • for instance, traffic transportation studies may require a change detection period of a few minutes.
    • You don't need daily images to study deforastration.
  • Temporal resolution should be held constant during change detection, if possible.

  • Images used for change detection should be acquired at comparable times.

    • for instance, minimize seasonal or tidal effects.

1. Data Processing

Image rectification

  • Geometric rectification may be needed to eliminate systematic and non-systematic distortions caused by different RS platforms

  • Radiometric correction may be needed to eliminate avoid radiometric distortions.

2. Change Detection

Volumne Change

  • Change of certain quanities over time
  • e.g. changes of NDVI (vegetation), impervious surface (urban), ice cap.

Spatial pattern Change

  • Changes of spatial pattern and arrangement
  • e.g. fragmentation, clustering, spatial autocorrelation.

2. Change Detection

Spatial relation change

  • Change of spatial or non-spatial relation with other variables
  • e.g. increasing deforastration near roads, increasing urbanization in higher elevation regions.

Transition Frequency

  • Frequencies of different types of changes (Markov Chain)
  • e.g. what is the most frequent land cover transition, probability of a certain type of transition

Visualizing change (Image comparison)

Visualizing change (Image comparison)

Visualizing change (Color-code changed pixels)

Visualizing change (Color-code changed pixels)

Visualizing change (Color-code changed pixels)

Animation

Google Earth Time Laps

  • Based on time series of Landsat images
  • Visually observe land cover changes over time.

https://earthengine.google.com/timelapse/

Quantitative Detection

Example: Land-Cover Change in Upper Barataria Basin Estuary, Louisiana, 1972-1992: Increases in Wetland Area

Nelson et al., 2002, Land-Cover Change in Upper Barataria Basin Estuary, Louisiana, 1972-1992: Increases in Wetland Area, Environmental Management, 29(5), 716-727 pdf

Study Area

Image time series

Changes of different land cover types

Rate of change

Transition matrix (in ha)

Transition matrix (in %)

Bottomland forest and swamp forest in Louisiana

New development

UAV

Building Change Detection using Unmanned Aerial Vehicle (UAV)

New development

UAV

Constructing and Comparing 3D building changes Captured by RGB-D camera

New development

High resolution images (e.g. Digital Globe)

  • 30m resolution in 30min frequency
  • Estimate Canadian's retailing industry by 'counting' cars in supermarkets' parking lots.
  • Estimate China's economy by 'counting' trucks parked in factories in southeastern China.

http://global.digitalglobe.com/interactive/30cm/

http://microsites.digitalglobe.com/30cm/

Lab Assignment 3

Download the assignment from https://git.io/vDRCs

Submission due on March 3